Method Of Image Quality Assessment Using Structural Similarity Based On Gradient

The results of image quality assessment must match visual effects of human visionsystem, but the existing assessment methods can not make them perfectly. Therefore thestandard of objective image quality assessment becomes a hot spot in the research field.Structure Similarity (SSIM）as a well-known objective image quality assessment isproposed by Zhou Wang. SSIM assumes that human visual perception is highly adapted forextracting structure information from a scene. Compared with PSNR，SSIM has a strongadvantage which has been proved in many different image quality assessments. However, dueto the HVS characteristics of the underlying visual are neglected, SSIM has no good resultsmatching to the subjective assessment. In this thesis，a modified SSIM method---GWSSIM isproposed, which is based on gradient-weighted by using structural similarities. The structurein the third part of SSIM is replaced by the gradient based calculation in this method. Thecomputer simulation results show that GWSSIM model is an effective method, which is betterthan the SIM model especially for the blurred image.